Supervised machine learning for social scientists

less than 1 minute read

Much of the work of me and my colleagues features prediction techniques that are not very (but increasingly) common in psychological research. Thus, much of the texts I write and discussions I have explain or revolve around these techniques. Often, social and behavioral scientists are surprised how easily they comprehend these techiques as they relate closely to methods that they already mastered. In this paper, we simply aimed to give any empirical scientist the insights they need to start using machine learning techniques themselves. That means we provide plenty of readable text, annotated R code, templates for model preregistration, and decision guides for data collection, modeling techniques, hyperparameters, accuracy metrics, and reporting. Our hope is that the paper and all the supplementary materials support work focussing on accurately predicting human behavior, which can be more challenging than most publications seem to suggest.

If you are interested, please read the postprint here: Publication in Social and Personality Psychology Compass